[httperf] How to use multiple clients generate high workload?

[Suchada Pakapongpan]

Because I use 1 client to generate 20 sessions/s.

#httperf --hog  --timeout=150 --client=0/1 --server=10.1.1.1 --port=80 --wsesslog=10000,0,input.txt --rate=20

my result has lot of fd-unavail error I guess the client was overloaded. Is it correct?

Errors: total 23895 client-timo 65 socket-timo 0 connrefused 0 connreset 20942

Errors: fd-unavail 2888 addrunavail 0 ftab-full 0 other 0

Then I use 2 clients to generate workload. Each client used this script:

#httperf --hog  --timeout=150 --client=0/2 --server=10.1.1.1 --port=80 --wsesslog=10000000,0,input.txt --rate=10
#httperf --hog  --timeout=150 --client=1/2 --server=10.1.1.1 --port=80 --wsesslog=10000000,0,input.txt --rate=10

I start that script in the same time for the result I sum the reply rate and average reply time.I’d like to know for my testing way. Is it correct?

And if I have to use more clients how can I generate the workload? (it’s hard to run multiple clients in the same time)
 
Thanks,
Cooky 		 	   		  
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://napali.hpl.hp.com/pipermail/httperf/attachments/20101117/19f98c25/attachment.htm